The invention relates to apparatus and methods for carrying out electrodeionization to purify water. Electrodeionization is a process for removing ions from liquids by sorption of these ions into a solid material capable of exchanging these ions for either hydrogen ions (for cations) or hydroxide ions (for anions) and simultaneous or later removal of the sorbed ions into adjacent compartments by the application of an electric field. (See Glueckauf, E., "Electro-Deionization Through a Packed Bed", December 1959, pp. 646-651, British Chemical Engineering for a background discussion.) The hydrogen and hydroxide ions needed to drive the ion exchange process are created by splitting of water molecules at the interface of anion and cation exchanging solids that contact each other in the orientation that depletes the contact zone of ions, when in the presence of an electric field. This orientation requires that the anion exchanging material face the anode and the cation exchanging material face the cathode. The created hydroxide ions enter the anion exchanging material, and the created hydrogen ions enter the cation exchanging material.
The electrodeionization process is commonly carried out in an apparatus consisting of alternating diluting compartments and concentrating compartments separated by anion permeable and cation permeable membranes. The diluting compartments are filled with porous ion exchanging solid materials through which the water to be deionized flows. The ion exchanging materials are commonly mixtures of cation exchanging resins and anion exchanging resins (e.g., U.S. Pat. No. 4,632,745), but alternating layers of these resins have also been described (e.g., U.S. Pat. Nos. 5,858,191 and 5,308,467). Ion exchanging materials consisting of woven and non-woven fibers have also been described. (e.g., U.S. Pat. Nos. 5,308,467 and 5,512,173). The compartments adjoining the diluting compartment into which the ions are moved by the applied electric field, called concentrating compartments, may be filled with ion exchanging materials or with inert liquid permeable materials. An assembly of one or more pairs of diluting and concentrating compartments, referred to as a "cell pair", is bounded on either side by an anode and a cathode which apply an electric field perpendicular to the general direction of liquid flow. Flow of water is provided past the anode and cathode.
The diluting compartments are each bounded on the anode side by an anion permeable membrane and on the cathode side by a cation permeable membrane. The adjacent concentrating compartments are each correspondingly bounded by a cation permeable membrane on the anode side and an anion permeable membrane on the cathode side. The applied electric field causes anions to move from the diluting compartment across the anion permeable membrane into the concentrating compartment nearer the anode and cations to move from the diluting compartment across the cation permeable membrane into the concentrating compartment nearer the cathode. The anions and cations become trapped in the concentrating compartments because the movement of anions toward the anode is blocked by a cation permeable membrane, and the movement of cations toward the cathode is blocked by an anion permeable membrane. A flow of water is set up to remove the ions from the concentrating compartments. The net result of the process is the removal of ions from the water stream flowing through the diluting compartments and their concentration in the water flowing through the concentrating compartments.
Electrodeionization (EDI) stacks frequently suffer from precipitation of calcium carbonate in the concentrating compartments as well as in the cathode compartment. (See AEDI and Membranes: Practical Ways to Reduce Chemical Usage when Producing High Purity Water, William E. Katz in Ultrapure Water, Vol. 16, No. 6 July/August 1999, pp 52-57). The consequence of this "scaling" is an increase in stack resistance, a drop in current density and eventually a sharp decrease in the purity of the product water. Vendors of EDI equipment suggest that the concentration of calcium in the feed to the EDI be limited to very low levels; e.g., less than 0.5 ppm. (U.S. Filter Literature No. US2006). While this concentration can be achieved when the electrodeionization apparatus is fed with reverse osmosis (RO) permeate from an RO system with new membranes, and the RO system is operating properly, the suggested values can be exceeded when these conditions do not hold.
In order for calcium carbonate to precipitate in solution the Langelier Saturation Index (LSI) has to be positive. In the cathode compartment the pH can be high enough for the LSI to be positive; precipitation of calcium carbonate is therefore to be expected under some circumstances. The LSI of RO permeates is always negative. Even in the EDI brine the concentrations of calcium and bicarbonate are so low that the LSI is still negative, at the prevailing pH. Thus, on the basis of consideration of LSI alone, one would not expect the precipitation of calcium carbonate that occurs within EDI concentrating compartments. This phenomenon is instead explainable based upon local conditions.
When a concentrating compartment from a Ascaled@ EDI stack is examined, the scale is observed on the anion membrane, predominantly halfway between the inlet and the outlet of the stack. This pattern can be explained on the basis of the mechanism by which an EDI stack removes weak acids like carbon dioxide and silica. At the pH of RO permeate, only a tiny fraction of silica is ionized, and a large fraction of the total inorganic carbon (TIC) is in the form of carbon dioxide. In order for silica and carbon dioxide to be removed by EDI, the feed solution needs to contact anion exchange resin in the diluting compartment, which is partly in the OH-- form, regenerated. Carbon dioxide and silica diffuse from solution into the partly regenerated anion resin and react with the OH.sup.- to form the HCO.sub.3.sup.-, CO.sub.3.sup.50 and HSiO.sub.3.sup.- anions which are moved, along with substantial amounts of OH.sup.-, by the applied voltage gradient, into the concentrating compartment. In order for the above mechanism to operate, the voltage drop in the diluting compartment has to be high enough, typically 2 to 3 volts, to regenerate some portion of the anion resin by the splitting of water into OH.sup.- and H.sup.+.
At the inlet portion of an EDI stack the extent of resin regeneration in the diluting compartment is low. Carbon dioxide and silica are therefore not removed in the front part of the stack. Toward the middle of the stack the concentrations of the ions in the feed water have dropped sharply and water splitting takes place. The resins are partly regenerated and the carbon dioxide and silica are removed. The pH of the solution on the concentrating side surface of the anion membrane is therefore very high; the concentration of CO.sub.3.sup..dbd. is also high, and the LSI can be positive at the concentration of calcium prevailing in the concentrate. (See U.S. Pat. No. 5,593,563). Calcium carbonate can therefore precipitate, as shown in FIG. 5. Note that the LSI within the bulk of the concentrate is still negative because the pH of the concentrate is virtually the same as that of the feed. The high pH at the surface of the anion membrane and the corresponding low pH at the surface of the cation membrane are boundary layer phenomena.
Toward the outlet of the stack virtually all of the anions have been removed. Although the concentrating side of the anion membrane is still at a very high pH, the concentration of CO.sub.3.sup..dbd. is so low that the LSI index is negative, and calcium carbonate does not precipitate. If it were not for the need to remove the weak acids by operating the EDI stack in a partly regenerated form, there would not be any problem with calcium carbonate precipitation. In order for EDI stacks to replace ion exchange beds, which remove these weak acids, EDI stacks must be operated in a partly regenerated form and consequently calcium carbonate precipitation is always a threat. The problem of calcium carbonate precipitation has been broadly recognized, and various suggestions have been made to deal with it. One approach is the periodic reversal of the role of the diluting and concentrating compartments with a simultaneous reversal of the polarity of the electrodes. (e.g., U.S. Pat. Nos. 4,956,071 and 5,558,753). Drawbacks of this approach include the production of low quality water during some parts of the operating cycle and the complexity and expense of the valving system needed to implement the process.
The special problem of calcium carbonate precipitation in the cathode compartment, exacerbated by the formation of hydroxide ions, has been dealt with by filling the cathode compartment with an electrically conductive medium. (e.g., U.S. Pat. No. 5,593,563). This is said to reduce the concentration of hydroxide ions at the surface of the electrode by distributing the current over a larger area and thus decreasing the degree of calcium carbonate supersaturation.
Calcium carbonate scaling can be prevented by reducing or eliminating any of the three prerequisites of scaling; calcium, carbon dioxide and bicarbonate or alkaline pH. The brute force chemical approaches--softening the RO feed or adding acid to the cathode compartment or to the concentrate--re-institute the very problems of chemical supply and waste disposal that EDI is designed to eliminate and are therefore fundamentally unacceptable. More acceptable approaches are the softening of the EDI feed, which has a much lower concentration of calcium than the RO feed, or the removal of carbon dioxide by air stripping. These approaches entail additional capital costs and operating expenses. It has also been suggested that the concentration of salts in the concentrating compartment be reduced by reducing the fraction of the feed water that is recovered as the pure water stream. This approach is fundamentally unacceptable because of its expense.